Skip to main content
Version: Next

Migrations

MikroORM has integrated support for migrations via umzug. It allows you to generate migrations based on the current schema difference.

info

To use migrations, you need to first install @mikro-orm/migrations package for SQL driver or @mikro-orm/migrations-mongodb for MongoDB, and register the Migrator extension in your ORM config.

mikro-orm.config.ts
import { Migrator } from '@mikro-orm/migrations'; // or `@mikro-orm/migrations-mongodb`

export default defineConfig({
  // ...
  extensions: [Migrator],
})

Since v5, migrations are stored without an extension.

By default, each migration will be executed inside a transaction, and all of them will be wrapped in one master transaction, so if one of them fails, everything will be rolled back.

Migration class​

Migrations are classes that extend Migration abstract class:

import { Migration } from '@mikro-orm/migrations';

export class Migration20191019195930 extends Migration {

  async up(): Promise<void> {
    this.addSql('select 1 + 1');
  }

}

To support undoing those changed, you can implement the down method, which throws an error by default.

Migrations are by default wrapped in a transaction. You can override this behaviour on per migration basis by implementing the isTransactional(): boolean method.

Configuration object and driver instance are available in the Migration class context.

You can execute queries in the migration via Migration.execute() method, which will run queries in the same transaction as the rest of the migration. The Migration.addSql() method also accepts instances of knex. Knex instance can be accessed via Migration.getKnex();

Working with EntityManager​

While the purpose of migrations is mainly to alter your SQL schema, you can as well use them to modify your data, either by using this.execute(), or through an EntityManager:

warning

Using the EntityManager in migrations is possible, but discouraged, as it can lead to errors when your metadata change over time, since this will depend on your currently checked out app state, not on the time when the migration was generated. You should prefer using raw queries in your migrations.

import { Migration } from '@mikro-orm/migrations';
import { User } from '../entities/User';

export class Migration20191019195930 extends Migration {

  async up(): Promise<void> {
    const em = this.getEntityManager();
    em.create(User, { ... });
    await em.flush();
  }

}

Initial migration​

This is optional and only needed for the specific use case, when both entities and schema already exist.

If you want to start using migrations, and you already have the schema generated, you can do so by creating so-called initial migration:

Initial migration can be created only if there are no migrations previously generated or executed.

npx mikro-orm migration:create --initial

This will create the initial migration, containing the schema dump from schema:create command. The migration will be automatically marked as executed.

Snapshots​

Creating new migration will automatically save the target schema snapshot into migrations folder. This snapshot will be then used if you try to create new migration, instead of using current database schema. This means that if you try to create new migration before you run the pending ones, you still get the right schema diff.

Snapshots should be versioned just like the regular migration files.

Snapshotting can be disabled via migrations.snapshot: false in the ORM config.

Configuration​

Since v5, umzug 3.0 is used, and pattern option has been replaced with glob.

migrations.path and migrations.pathTs works the same way as entities and entitiesTs in entity discovery.

await MikroORM.init({
  // default values:
  migrations: {
    tableName: 'mikro_orm_migrations', // name of database table with log of executed transactions
    path: './migrations', // path to the folder with migrations
    pathTs: undefined, // path to the folder with TS migrations (if used, you should put path to compiled files in `path`)
    glob: '!(*.d).{js,ts}', // how to match migration files (all .js and .ts files, but not .d.ts)
    transactional: true, // wrap each migration in a transaction
    disableForeignKeys: true, // wrap statements with `set foreign_key_checks = 0` or equivalent
    allOrNothing: true, // wrap all migrations in master transaction
    dropTables: true, // allow to disable table dropping
    safe: false, // allow to disable table and column dropping
    snapshot: true, // save snapshot when creating new migrations
    emit: 'ts', // migration generation mode
    generator: TSMigrationGenerator, // migration generator, e.g. to allow custom formatting
  },
})

Running migrations in production​

In production environment you might want to use compiled migration files. Since v5, this should work almost out of box, all you need to do is to configure the migration path accordingly:

import { MikroORM, Utils } from '@mikro-orm/core';

await MikroORM.init({
  migrations: {
    path: 'dist/migrations',
    pathTs: 'src/migrations',
  },
  // or alternatively
  // migrations: {
  //   path: Utils.detectTsNode() ? 'src/migrations' : 'dist/migrations',
  // },
  // ...
});

This should allow using CLI to generate TS migration files (as in CLI you probably have TS support enabled), while using compiled JS files in production, where ts-node is not registered.

Using custom MigrationGenerator​

When you generate new migrations, MigrationGenerator class is responsible for generating the file contents. You can provide your own implementation to do things like formatting the SQL statement.

import { TSMigrationGenerator } from '@mikro-orm/migrations';
import { format } from 'sql-formatter';

class CustomMigrationGenerator extends TSMigrationGenerator {

  generateMigrationFile(className: string, diff: { up: string[]; down: string[] }): string {
    const comment = '// this file was generated via custom migration generator\n\n';
    return comment + super.generateMigrationFile(className, diff);
  }

  createStatement(sql: string, padLeft: number): string {
    sql = format(sql, { language: 'postgresql' });
    // a bit of indenting magic
    sql = sql.split('\n').map((l, i) => i === 0 ? l : `${' '.repeat(padLeft + 13)}${l}`).join('\n');

    return super.createStatement(sql, padLeft);
  }

}

await MikroORM.init({
  // ...
  migrations: {
    generator: CustomMigrationGenerator,
  },
});

Using via CLI​

You can use it via CLI:

npx mikro-orm migration:create   # Create new migration with current schema diff
npx mikro-orm migration:up       # Migrate up to the latest version
npx mikro-orm migration:down     # Migrate one step down
npx mikro-orm migration:list     # List all executed migrations
npx mikro-orm migration:check    # Check if schema is up to date
npx mikro-orm migration:pending  # List all pending migrations
npx mikro-orm migration:fresh    # Drop the database and migrate up to the latest version

To create blank migration file, you can use npx mikro-orm migration:create --blank.

For migration:up and migration:down commands you can specify --from (-f), --to (-t) and --only (-o) options to run only a subset of migrations:

npx mikro-orm migration:up --from 2019101911 --to 2019102117  # the same as above
npx mikro-orm migration:up --only 2019101923                  # apply a single migration
npx mikro-orm migration:down --to 0                           # migrate down all migrations

To run TS migration files, make sure you have ts-node installed in your project, the CLI will register it automatically since v6.3.

For the migration:fresh command you can specify --seed to seed the database after migrating.

npx mikro-orm migration:fresh --seed              # seed the database with the default database seeder
npx mikro-orm migration:fresh --seed=UsersSeeder  # seed the database with the UsersSeeder

You can specify the default database seeder in the orm config with the key config.seeder.defaultSeeder

Using the Migrator programmatically​

Or you can create a simple script where you initialize MikroORM like this:

./migrate.ts
import { MikroORM } from '@mikro-orm/core';
import { Migrator } from '@mikro-orm/migrations';

(async () => {
  const orm = await MikroORM.init({
    extensions: [Migrator],
    dbName: 'your-db-name',
    // ...
  });

  const migrator = orm.getMigrator();
  await migrator.createMigration(); // creates file Migration20191019195930.ts
  await migrator.up(); // runs migrations up to the latest
  await migrator.up('name'); // runs only given migration, up
  await migrator.up({ to: 'up-to-name' }); // runs migrations up to given version
  await migrator.down(); // migrates one step down
  await migrator.down('name'); // runs only given migration, down
  await migrator.down({ to: 'down-to-name' }); // runs migrations down to given version
  await migrator.down({ to: 0 }); // migrates down to the first version

  await orm.close(true);
})();

Then run this script via ts-node (or compile it to plain JS and use node):

$ ts-node migrate

Providing transaction context​

In some cases, you might want to control the transaction context yourself:

await orm.em.transactional(async em => {
  await migrator.up({ transaction: em.getTransactionContext() });
});

Importing migrations statically​

If you do not want to dynamically import a folder (e.g. when bundling your code with webpack) you can import migrations directly. You can do that with an explicit migration name or the implicit filename as migration name.

import { MikroORM } from '@mikro-orm/core';
import { Migrator } from '@mikro-orm/migrations';
import { Migration20191019195930 } from '../migrations/Migration20191019195930.ts';
import { Migration20191019195931 } from '../migrations/Migration20191019195931.ts';

await MikroORM.init({
  extensions: [Migrator],
  migrations: {
    migrationsList: [
      // explicit migration name
      {
        name: 'CustomMigrationName',
        class: Migration20191019195930,
      },
      // implicit migration name
      Migration20191019195931
    ],
  },
});

With the help of webpack's context module api you can dynamically import the migrations making it possible to import all files in a folder.

import { MikroORM } from '@mikro-orm/core';
import { Migrator } from '@mikro-orm/migrations';
import { basename } from 'path';

const migrations = {};

function importAll(r) {
  r.keys().forEach(
    (key) => (migrations[basename(key)] = Object.values(r(key))[0])
  );
}

importAll(require.context('../migrations', false, /\.ts$/));

const migrationsList = Object.keys(migrations).map((migrationName) => ({
  name: migrationName,
  class: migrations[migrationName],
}));

await MikroORM.init({
  extensions: [Migrator],
  migrations: {
    migrationsList,
  },
});

Using custom migration names​

Since v5.7, you can specify a custom migration name via --name CLI option. It will be appended to the generated prefix:

# generates file Migration20230421212713_add_email_property_to_user_table.ts
npx mikro-orm migration:create --name=add_email_property_to_user_table

You can customize the naming convention for your migration file by utilizing the fileName callback, or even use it to enforce migrations with names:

migrations: {
  fileName: (timestamp: string, name?: string) => {
    // force user to provide the name, otherwise you would end up with `Migration20230421212713_undefined`
    if (!name) {
      throw new Error('Specify migration name via `mikro-orm migration:create --name=...`');
    }

    return `Migration${timestamp}_${name}`;
  },
},
Warning

When overriding the migrations.fileName strategy, keep in mind that your migration files need to be sortable, you should never start the filename with the custom name option as it could result in wrong order of execution.

MongoDB support​

Support for migrations in MongoDB has been added in v5.3. It uses its own package: @mikro-orm/migrations-mongodb, and should be otherwise compatible with the current CLI commands. Use this.driver or this.getCollection() to manipulate with the database.

Transactions​

The default options for Migrator will use transactions, and those impose some additional requirements in mongo, namely the collections need to exist upfront, and you need to run a replicaset. You might want to disable transactions for migrations: { transactional: false }.

await this.driver.nativeDelete('Book', { foo: true }, { ctx: this.ctx });

You need to provide the transaction context manually to your queries, either via the ctx option of the driver methods, or via the MongoDB session option when using the this.getCollection() method.

await this.getCollection('Book').updateMany({}, { $set: { updatedAt: new Date() } }, { session: this.ctx });

Migration class​

Example migration in mongo:

import { Migration } from '@mikro-orm/migrations-mongodb';

export class MigrationTest1 extends Migration {

  async up(): Promise<void> {
    // use `this.getCollection()` to work with the mongodb collection directly
    await this.getCollection('Book').updateMany({}, { $set: { updatedAt: new Date() } }, { session: this.ctx });

    // or use `this.driver` to work with the `MongoDriver` API instead
    await this.driver.nativeDelete('Book', { foo: true }, { ctx: this.ctx });
  }

}

Limitations​

MySQL​

There is no way to rollback DDL changes in MySQL. An implicit commit is forced for those queries automatically, so transactions are not working as expected.

MongoDB​

  • no nested transaction support
  • no schema diffing
  • only blank migrations are generated

Debugging​

Sometimes the schema diffing might not work as expected and will produce unwanted queries. Often this is a problem with how you set up the columnType or default/defaultRaw options of your properties. You can use the MIKRO_ORM_CLI_VERBOSE environment variable to enable verbose logging of the CLI. This, in turn, enables logging of both the underlying queries used to extract the current schema and the logs in the SchemaComparator, which should help you understand why the ORM sees two columns as different and what particular options are different.

Debugging issues with migrations is easier when you use schema:update, as you skip the Migrator layer on top of it and test the actual layer where those problems occur.

$ MIKRO_ORM_CLI_VERBOSE=1 npx mikro-orm schema:update --dump